E-commerce Simulation + AI Shopping Assistant

Overview

This project simulates a full, end-to-end e-commerce experience using real product metadata from the Kaggle H&M dataset. I built the frontend as a responsive storefront with Next.js App Router and TypeScript modelling real online e-commerce shops with all the features you would expect: a product catalog, a search bar to query products, an add to cart feature, an onboarding feature as well as a chatbot to help guide the user to the best-fit products. On the backend, I used FastAPI with SQLite to serve product data through a set of REST endpoints. This keeps the system realistic: the frontend isn't mock data—it fetches products, details, and similar items dynamically, like a real commerce stack. The standout feature is an AI shopping assistant embedded in the site. Instead of being a generic chatbot, it's grounded in the actual catalog: the assistant receives relevant product context from the backend (based on the user's query or browsing state) and uses that context to answer questions and recommend items without inventing products. If something isn't present in the dataset (like sizing/fit), the assistant explicitly says it's unavailable.

Highlights

• Search + filtering UX (category browsing, fast search, filters)
• Product detail pages + similar items (catalog-based recommendations)
• Cart + checkout flow (add/remove/update quantities, responsive UI states)
• AI shopping assistant grounded in catalog context (no hallucinated products)